Figure 1.

Coronary Ultrafast Doppler Angiography Acquisition Sequence and Processing

(A) Conventional ultrasound imaging in real time was used to select the heart section of interest (mid parasternal short-axis view here). The region of interest of the anterior wall is outlined in yellow. (B) Ultrafast acquisition of the anterior wall region of interest and corresponding electrocardiogram. A total of 1,000 frames were typically acquired in diastole over 0.5 s. (C) An adaptive spatiotemporal filter was used to separate coronary flow from tissue clutter through diastole using a sliding window analysis. Using a singular value decomposition on sliding short ensembles of frames (<50), we separated tissue signal (gray-scale image) contained in the highest eigenvectors from blood signal contained in the lowest eigenvectors (red-scale image). Our filtering method relies on the fact that tissue is highly echogenic and exhibits cohesive motion throughout the image plane, whereas coronary blood signal is weakly echogenic and has a random motion nature, because erythrocytes are freely floating within vessels.